Abstract: "We study whether image preferences in isolation from strategic considerations, namely purely hedonic image concerns, can motivate prosocial behavior and whether this audience effect is mediated by the number of observers. Answers to related questions from the extant experimental literature are often mixed or influenced by multiple mechanisms evoked by the context at hand or design employed. We employ an experiment involving a dictator game with a charity receiver and a binary choice with unambiguous social valence. Choices are observed by an anonymous, passive, and external audience whose size varies across treatments. Our simple experimental design allows us to isolate purely hedonic image concerns about appearing altruistic from strategic considerations and other confounding features of alternative designs. We find that donations rise by 10.2 percentage points on average when audiences are present, with every observer increasing the probability of donating by an estimated 2.12 percentage points. We provide evidence that the size of the audience also matters."